CanItRun Logocanitrun.

DeepSeek R1 671B vs Qwen3 235B-A22B (MoE)

Side-by-side VRAM requirements, benchmark scores, and GPU compatibility for local AI inference.

Quick verdict

Qwen3 235B-A22B (MoE) is more hardware-efficient: it needs 162.1 GB at its Q4_K_M build vs 458.3 GB for DeepSeek R1 671B's Q4_K_M, fitting on 16 GPUs natively.

VRAM at each quantization (8k context)

FP32
DeepSeek R1 671B
3006.7 GB
Qwen3 235B-A22B (MoE)
1054.6 GB
BF16
DeepSeek R1 671B
1503.6 GB
Qwen3 235B-A22B (MoE)
528.2 GB
FP16
DeepSeek R1 671B
1503.6 GB
Qwen3 235B-A22B (MoE)
528.2 GB
Q8_0
DeepSeek R1 671B
799.4 GB
Qwen3 235B-A22B (MoE)
281.6 GB
Q6_K
DeepSeek R1 671B
617.6 GB
Qwen3 235B-A22B (MoE)
217.8 GB
Q5_K_M
DeepSeek R1 671B
535.7 GB
Qwen3 235B-A22B (MoE)
189.2 GB
Q4_K_M
DeepSeek R1 671B
458.3 GB
Qwen3 235B-A22B (MoE)
162.1 GB
Q3_K_M
DeepSeek R1 671B
362.1 GB
Qwen3 235B-A22B (MoE)
128.4 GB
Q2_K
DeepSeek R1 671B
286.9 GB
Qwen3 235B-A22B (MoE)
102.0 GB
NVFP4
DeepSeek R1 671B
376.3 GB
Qwen3 235B-A22B (MoE)
133.4 GB
QuantDeepSeek R1 671BQwen3 235B-A22B (MoE)Diff
FP323006.7 GB1054.6 GB+185%
BF161503.6 GB528.2 GB+185%
FP161503.6 GB528.2 GB+185%
Q8_0799.4 GB281.6 GB+184%
Q6_K617.6 GB217.8 GB+183%
Q5_K_M535.7 GB189.2 GB+183%
Q4_K_M458.3 GB162.1 GB+183%
Q3_K_M362.1 GB128.4 GB+182%
Q2_K286.9 GB102.0 GB+181%
NVFP4376.3 GB133.4 GB+182%

Diff is DeepSeek R1 671B relative to Qwen3 235B-A22B (MoE). Green = lower VRAM (fits more GPUs).

Model specifications

SpecDeepSeek R1 671BQwen3 235B-A22B (MoE)
OrgDeepSeekAlibaba
Parameters671B235B
ArchitectureMoE (37B active)MoE (22B active)
Context125k tokens128k tokens
Modalitiestexttext
LicenseMITApache 2.0
CommercialYesYes
Released2025-01-202025-04-29
GPUs (native)2 / 11916 / 119

Benchmark scores

BenchmarkDeepSeek R1 671BQwen3 235B-A22B (MoE)
MMLU-Pro85.084.4
GPQA Diamond71.5N/A
IFEval83.3N/A
MATH97.3N/A
LiveCodeBench65.9N/A

Green = higher score (better). N/A = not yet available. ~ = inherited from a base model, not independently reported for that release itself.

GPUs that run only DeepSeek R1 671B(0)

Every GPU that runs DeepSeek R1 671B also runs Qwen3 235B-A22B (MoE).

GPUs that run only Qwen3 235B-A22B (MoE)(14)

GPUs that run both natively(2)

Which should you use?

Choose DeepSeek R1 671B if:
  • • You want maximum capability and have a 459 GB+ GPU
  • • Benchmark quality matters: scores 85.0 vs 84.4 on MMLU-Pro
Choose Qwen3 235B-A22B (MoE) if:
  • • You have limited VRAM: it's a smaller model needing 162.1 GB vs 458.3 GB
  • • Long context matters: it supports 128k tokens vs 125k
  • • It's the newer release (2025-04-29 vs 2025-01-20); check the benchmark table above for what actually improved

Frequently asked questions

Which is better, DeepSeek R1 671B or Qwen3 235B-A22B (MoE)?
DeepSeek R1 671B has 671B parameters vs 235B for Qwen3 235B-A22B (MoE), so DeepSeek R1 671B is the larger model. Qwen3 235B-A22B (MoE) is more hardware-efficient, needing 162.1 GB at its Q4_K_M build vs 458.3 GB for DeepSeek R1 671B's Q4_K_M. Qwen3 235B-A22B (MoE) runs on more GPUs natively (16 vs 2). On MMLU-Pro, DeepSeek R1 671B scores higher (85.0 vs 84.4).
How much VRAM does DeepSeek R1 671B need vs Qwen3 235B-A22B (MoE)?
At 8k context, DeepSeek R1 671B needs approximately 458.3 GB of VRAM at its Q4_K_M build, while Qwen3 235B-A22B (MoE) needs 162.1 GB at its Q4_K_M build. At the largest build each ships, DeepSeek R1 671B requires 1503.6 GB (FP16) vs 528.2 GB (FP16) for Qwen3 235B-A22B (MoE).
Can you run DeepSeek R1 671B on the same GPUs as Qwen3 235B-A22B (MoE)?
Yes, 2 GPUs can run both natively in VRAM, including Apple M5 Ultra (512GB), Apple M3 Ultra (512GB). However, no GPU can run DeepSeek R1 671B without also fitting Qwen3 235B-A22B (MoE), and 14 GPUs can run Qwen3 235B-A22B (MoE) but not DeepSeek R1 671B.
What is the difference between DeepSeek R1 671B and Qwen3 235B-A22B (MoE)?
DeepSeek R1 671B has 671B parameters (37B active, MoE) with a 125k context window. Qwen3 235B-A22B (MoE) has 235B parameters (22B active, MoE) with a 128k context window. Licensing differs: DeepSeek R1 671B is MIT while Qwen3 235B-A22B (MoE) is Apache 2.0.
Which model fits in 24 GB of VRAM, DeepSeek R1 671B or Qwen3 235B-A22B (MoE)?
Neither fits in 24 GB: DeepSeek R1 671B needs 458.3 GB at Q4_K_M and Qwen3 235B-A22B (MoE) needs 162.1 GB at Q4_K_M. Both require a multi-GPU server with 459 GB+ of combined VRAM.
Full DeepSeek R1 671B page →Full Qwen3 235B-A22B (MoE) page →Check your hardware →